“X” marks the spot: Mining the gold in CasX for gene editing

Abstract

In this issue of Molecular Cell, Tsuchida et al. (2022) present a successful structure-guided effort in improving genome-editing efficiencies of CRISPR-CasX from Deltaproteobacteria (DpbCasX) and Planctomycetes (PlmCasX). Engineered variants that stabilize the active conformational state improved the catalytic efficiency by ~10–20 fold in vitro and mean-editing efficiency by ~2–3 fold in human cells.

The discovery and adoption of clustered regularly interspaced short palindromic repeat (CRISPR)-Cas (CRISPR-associated) DNA endonucleases have opened doors to explosive applications in genome manipulation and clinical interventions (Jiang and Doudna, 2017; Zhang, 2019). The first characterized of these enzymes is Cas9 (Jinek et al., 2012) that was soon followed by Cas12a, or Cpf1 (Zetsche et al., 2015). Both Cas9 and Cpf1 achieve RNA-guided DNA double-stranded breaks and have been shown to be highly effective in various organisms. The continued and near-universal use of Cas9 and Cpf1 has identified several outstanding questions that must be addressed for improved capabilities: (1) how cleavage at unintended sites (off-targets) is avoided; (2) how cleavage at specific sites is expanded; and (3) how the catalytic efficiency is improved. One way to address these questions is through mining the enormous repertoire of microorganisms for the enzymes with desired properties, which, however, does not always control the yield. The other is to engineer the existing enzymes based on the knowledge of their three-dimensional structures, either by library-based selection or rational designs. Through an extraordinary effort, Tsuchida et al. in this issue of Molecular Cell report successful structure-based engineering of a new class of CRISPR-Cas enzyme, CasX, to have a remarkable efficiency closely matching that of the commonly used Streptococcus pyogenes Cas9 (SpyCas9) (Tsuchida et al., 2022). CasX is one of the smallest gene-editing enzymes and thus more amenable for convenient delivery. This work exemplifies a path for improving catalytic efficiency of the similar enzymes.

Tsuchida et al. employed a highly tractable system that allowed them to systematically isolate individual contributing factors. Two homologous CasX enzymes, from Deltaproteobacteria (DpbCasX) and Planctomycetes (PlmCasX), were identified in 2017 from metagenomic sequencing based on the presence of the catalytic RuvC domain (Burstein et al., 2017). Subsequently, the biochemical and genome-editing properties of the reconstituted DpbCasX and PlmCasX and the structure of DpbCasX were characterized in 2019 (Liu et al., 2019). Surprisingly, despite the 68% sequence identity and the shared single-guide RNA (sgRNA), they differ noticeably in stability and DNA cleavage activities. In this work, Tsuchida et al. determined the structures of the wild-type and a variant of PlmCasX by electron cryomicroscopy (cryoEM). The complete structural and biochemical data of the two homologous enzymes guided the engineering efforts.

Tsuchida et al. first showed that elements stabilizing the active structure contribute to the catalytic efficiency. By comparing the DpbCasX and the PlmCasX structures, the authors identified key differences in three loop regions (regions 1–3, or R1–R3). In their shared active conformation, R1 stabilizes the scaffold stem of the sgRNA, R2 interacts with the region near the protospacer adjacent motif (PAM), and R3 stabilizes the DNA strand placed near the RuvC active site (Figure 1). Consistent with the functional assignment of these loops, the complex formed by PlmCasX, which lacks a full R1 loop, contains a large class of particles with disordered Helical-II domain that harbors R1. The authors then constructed respective variants of DpbCasX and PlmCasX that bear the loops absent from their wild-type forms and tested their in vitro cleavage and in vivo editing activities. Adding the scaffold stem-binding loop to PlmCasX (PlmCasX-R1) that is only present in DpbCasX (R1) improved its in vitro DNA cleavage activity by 3-fold. Similarly, transferring a loop to DpbCasX (DbpCasX-R3) that is only present in PlmCasX (R3) improved its activity by 1.6 fold. These results suggest that modifications of the factors that stabilize the active conformation improves the catalytic efficiency.

Figure 1. Schematic representation of the chimeric CasX constructs and their activity results.

Thick lines indicate regions (R1, R3, and RNA hinge) with significant contribution to improved activities. Key elements are labeled: motion of the scaffold stem around the hinge (depicted by a double arrow), target strand (TS), nontarget strand (NTS), and protospacer adjacent motif (PAM). Triple helix RNA structure is not depicted for simplicity.

Tsuchida et al. also discovered that elements that drive conformational dynamics are critical to catalytic efficiency. Classification and analyzing the population of the substrate-bound DpbCasX and PlmCasX particles revealed two sequential steps of reactions: (1) loading and cleaving the non-target DNA strand (State I) and (2) loading and cleaving the target strand (State II) (Liu et al., 2019; Tsuchida et al., 2022). Transition from State I to State II is facilitated by engagement of Helical-II domain with the scaffold stem of sgRNA, the rate of which may control the overall rate of cleavage. Given that the Helical-II domain (and the stem-binding R1 loop) of DpbCasX is mostly engaged but that of PlmCasX is not with the scaffold stem, authors hypothesize that increasing the flexibility of the scaffold stem can drive its conformational transition. They inserted a nucleotide near the tightly curved triple helix region and concurrently converted a GC pair to UU pair (sgRNAv2) at the rotation hinge of the scaffold stem (Figure 1).The resulting effect is remarkable: both Dpb CasX (DpbCasX-sgRNAv2) and PlmCasX (PlmCasX-sgRNAv2) showed elevated editing activities by 5.6 and 11 fold, respectively, from the use of wild-type sgRNAv1. When the engineered protein loops were combined with the sgRNAv2 construct (DbpCasX-R3-sgRNAv2 and PlmCasX-R1-sgRNAv2), in vitro cleavage kinetics improved astonishingly by ~10–20 fold and gene editing in mammalian cells improved by ~2–3 fold. Satisfactorily, cryo-EM structures of PlmCasX-sgRNAv2 bound with DNA substrate captured this stabilized form and thus supported the improved activity through conformational stabilization.

Other studies targeted to improve Cas9 efficiency also showed that stabilization of the active conformation is an effective strategy. Hand et al., by using a bacterial selection assay, identified mutations in the hinge supporting HNH domain rotation in SpyCas9 that improved gene knockin efficiency by 2 fold in HCT116 cells (Hand et al., 2021). Strecker et al. successfully converted Bacillus hisashii Cas12b (BhCas12b) from primarily nicking to cleaving dsDNA by introducing mutations that stabilize target strand binding (Strecker et al., 2019). These examples suggest that the strategies identified in CasX studies are generally applicable to other CRISPR-Cas enzymes.

Several important questions remain unanswered that require additional studies. There seem to be a discrepancy in fold increase between the in vitro and in vivo assayed activities, suggesting that the rate of DNA cleavage in vitro is not equal to that in cells. Identification of the inhibitory factors in cells can further remove barriers to improve editing efficiencies. Additionally, many Cas12 enzymes are known to have the collateral DNase activities, which form the basis for constructing Cas12-based nucleic acid detection methods (Freije and Sabeti, 2021). However, Tsuchida et al. show that CasX and its variants, though possessing the same RuvC domain, have significantly lower collateral DNase activity. What is the basis for the differential control of RuvC in different Cas enzymes? Answers to these questions wait for further characterization of CRISPR-Cas enzyme structures at high resolutions.